There is a very high incidence of treatment resistance in bipolar illness. Some two thirds of patients in academic centers remain highly symptomatic despite aggressive pharmacotherapy. Days depressed exceed days manic by a factor of three. This large pool of treatment-refractory patients has been the focus of study of the Branch both to better understand the pathophysiological mechanisms in recurrent unipolar and bipolar disorders and develop new therapeutic modalities. With the current availability of a series of treatment options, finding clinical and biological markers of which patients respond best to a given treatment, is a critical need for the field. For example, a double-blind, randomized trial of six weeks of treatment with an agent that enhances inhibitory GABAergic function (gabapentin, GPN), versus one that decreases excitatory glutamatergic function (lamotrigine, LTG), versus placebo, found significant benefit of LTG (20/39 or 51% response rate) over GPN (11/40 or 28%) and placebo (8/28 or 21%). Correlates of lamotrigine response included male gender, bipolarity, fewer prior clinical trials, and fewer prior hospitalizations for depression (Obrocea et al). Preliminary evidence suggests that a baseline pattern of hypo-perfusion on 0-15 PET scans was associated with clinical response. In relationship to the assessment of possible predictors of clinical response, depressed patients with global hypermetabolism on PET, especially in the left insula, are more likely to be responsive to carbamazepine (N=26), while those with the more classic pattern of frontal and left insula hypometabolism are more likely to be responsive to the dihydropyridine L-type calcium channel blocker nimodipine (Ketter et al). We have found that nimodipine increases somatostatin in cerebrospinal fluid (CSF) and those with lower CSF somatostatin levels at baseline are more likely to respond clinically to nimodipine (Frye et al). The Branch is now analyzing a series of studies of the use of repeated transcranial magnetic stimulation (rTMS) of the brain for the treatment of depression that we have helped pioneer. Several patients responded in a pilot study of the first high frequency (20Hz) stimulation at 80% of motor threshold of left frontal cortex. A double-blind, randomized, crossover trial then indicated significant antidepressant effects of active 20 Hz rTMS for two weeks compared with the sham (George et al). The next study assessed the differential responsivity to low-frequency (1 Hz) vs. higher frequency (20 Hz) rTMS vs. sham stimulation over left frontal cortex at 80% of motor threshold (MT). This study found differential clinical and metabolic responses within the same patient to these different frequencies. Moreover, those with a pattern of baseline hypometabolism tend to respond to the 20 Hz stimulation, while those with baseline patterns of hypermetabolism are more likely to respond to the 1 Hz stimulation (Kimbrell et al). Because the incidence and magnitude of clinical responsivity was not adequate for many patients, a fourth rTMS study using higher intensities (100% of MT) was conducted. This study replicated the findings of differential responsivity within individual patients, and strikingly, revealed that 20Hz stimulation increased 0-15 blood flow in a long lasting fashion, while 1 Hz rTMS decreased it (Speer et al). A further controlled study in normal volunteers has confirmed that 1 Hz rTMS over frontal cortex induces relative decrements in bilateral frontal and striatal metabolism on PET. Two other studies in normal volunteers showed the different patterns of activation on PET of 1Hz rTMS delivered over motor compared with prefrontal cortex (Speer et al). Together these studies in patients and volunteers validate the different physiological effects of high vs. low frequency rTMS and area of brain stimulated. The most recent rTMS study used higher intensities of rTMS stimulation (110% MT) for a longer time (3 weeks) in an attempt to increase the response rate. This study randomized patients to 20 Hz vs. 1Hz vs. sham stimulation, and results and regional brain activity correlates of response are currently being analyed. Thus, a number of promising and mechanistically novel treatment approaches have been prioneered in the Branch and the current effort is aimed at defining better optimal parameters for rTMS response and defining clinical and neurobiological markers of individual responsiveness.